No currently available single test of fetal well-being is a universal, infallible clinical tool that stands apart from the more general context of the total pregnancy. It is tempting to place a substantial emphasis on the results of a test when medical decision-making is needed. Perhaps some sense of proportion should be invoked at this point, beginning with the recognition that fetal heart rate data, regardless of how they are obtained, provide a limited window to view the fetus and its environment. The fact remains that the relatively low incidence of real fetal compromise and subsequent poor outcomes biases the efficacy of the NST and other well-being tools. Most high-risk populations have a prevalence of poor outcomes of approximately 10 per cent; consequently, the target population for any assessment test is quite small. One must be reassured by the extremely large reported experience of Platt et al that nonstress testing is clinically useful in the overwhelming majority of cases. The NST appears to be an excellent predictor of the healthy fetus. The test is very good at predicting the fetus who does not require acute or premature obstetric intervention: it thereby prevents pregnancies from being subjected to unnecessary iatrogenic risks and from incurring tremendous medical and emotional costs. Can the NST be improved? I think that there is still considerable opportunity for making this test more effective and accurate. Our experience supports the concepts of standardized testing conditions, the use of baseline data in addition to reactivity alone, the prolongation of test sessions to achieve a more representative view of the fetus, and ultimately an automated analysis that will aid the clinician in making more reliable assessments of the fetal heart rate data present in the actual tracing. It would be useful to see the NST better integrated into management schemes that incorporate other assessment techniques as well as pertinent clinical data. The tendency to view the test as a "stand alone" modality has underscored weaknesses inherent in all surveillance methods that have a naturally limited perspective on complex developmental and adaptive problems. There is also a need for the development of academic standards so that collected experiences can be directly compared and communication of testing results made less ambiguous. These are some of the goals that should be set for the future and whose realization will effect better care for the unique population served by obstetricians.